Circumference of a Circle

$C = 2\pi r$ looks like a definition of circumference, but read it the other way and it is the definition of $\pi$. Take any circle at all — a coin, a grain silo, the equator — measure the distance around and divide by the distance across, and you get the same number every time. That constancy is not obvious and it is not free; it is a genuine property of flat space, and it fails on a sphere, where a circle drawn around the north pole has a circumference of less than $\pi$ times its diameter measured along the surface. On paper it holds, so one number serves every circle, and $C = \pi d$ is the same statement written for the measurement you actually take.

A worked instance: a 3-inch nominal steel pipe has an outside diameter of 88.9 mm, so a strap around it reads $\pi \times 88.9 = 279$ mm. Run it backwards and you have the field trick — a flexible tape around any pipe, divided by $\pi$, gives the outside diameter without ever getting calipers onto it. A 700c bicycle wheel is about 2.10 m around, which is how a cycle computer turns wheel revolutions into distance, and why entering the wrong tyre size quietly skews every ride you record.

About $\pi$ itself. Archimedes bracketed it around 250 BC by squeezing a circle between an inscribed and a circumscribed 96-sided polygon, getting $223/71 < \pi < 22/7$ — correct to two decimals and honest about its own uncertainty, which was remarkable work for the third century BC. It cannot be written down exactly. Lambert proved in 1761 that $\pi$ is irrational, so no fraction equals it and no decimal expansion ever repeats, and Lindemann proved in 1882 that it is transcendental, meaning it satisfies no polynomial equation with whole-number coefficients. That second result is what finally killed squaring the circle, a construction problem people had chased for two thousand years.

Two places this goes wrong. The first, again, is radius against diameter: the formula wants $r$, you measured $d$, and forgetting to halve doubles your answer. Circumference is more forgiving than area here — the error is a factor of two rather than four — but it is still the wrong number of metres of insulation. The second is treating $22/7$ and $3.14$ as interchangeable with $\pi$. The first is 0.04% high, the second 0.05% low, which is invisible on a fence line and unacceptable on a machined bore. Finally, notice that circumference scales linearly while area scales as the square: double the pipe and you use twice the lagging but four times the flow area. Wrapping a large tank is cheap per litre stored, which is the same reason big things lose heat more slowly than small ones.